|
PRODUCTION AND PROCESS ANALYTICS"Everybody has data, but only those that efficiently convert their data into useful actions reap the benefits"
|
SOFTWARE AND SERVICES TO ANALYZE AND IMPROVE PRODUCTION PROCESSES
|
Our analytics services are hired by the management, process engineers or the laboratory group of our customers to extract actionable information from production data or to investigate production issues. We have relevant expertise in several areas:
Click here to learn how to start using data to create value with our Introduction to data science and brainstorming workshop & Exploratory assessment and pre-study |
|
Statistical Analysis
- Design of Experiments (DOE): identify the most efficient way to investigate theories or problems
- Measurement System Analysis (MSA): determine gauge repeatability and reproducibility of measurement equipment and methods
- Power analysis: calculate the minimal sample size such that one can be reasonably likely to detect an effect of a given size
- Analysis of variance (ANOVA)
- Investigation of correlations and causal relationships
- ...
Machine Learning
Machine learning is a vast field of techniques and competences to let computers learn useful information from data.
Benefits:
From an engineering point of view to learn parameters from data has a big advantage: no assumptions about the operation or structure of the observed system are necessary. Moreover, because these models are computationally very powerful, they succeed in capturing complex non-linear properties of the underlying systems which are very difficult to model using standard engineering approaches.
Application domains:
Machine Learning is, due to its versatility, applied in a large number of industrial settings and to various application areas. These include but are not limited to:
- computer vision
- robot locomotion
- soft sensor development
- prediction and forecasting
- classification of products
- computer vision
- robot locomotion
- soft sensor development
- prediction and forecasting
- classification of products
Software development to support experimentation and analysis
In addition to performing data analysis ourselves, we also develop custom software tools that allow our customers to do data analysis or build computer models themselves. In this way, our customers can increase the efficiency of their team, or realize new capabilities.
Some reference Projects
- We have performed a Measurement System Analysis for Eftec NV. In this project we did a gauge R&R study according to the automotive guidelines as described in AIAG MSA 4th edition to determine the repeatability and reproducibility of rheology measurements.
- We have designed and co-developed for the LT3 group of the University of Ghent a software tool that improves the efficiency of their research. This toolbox is called "Gallop", which stands for "Genetic Algorithms for Linguistic Learner Optimization". The tool launches experiments on a cluster of computers to find the optimal feature selection and the optimal learning parameters.
- We have identified the production variables that have unusual values and/or cause suboptimal production for several paper manufacturers and a copper mine, and found and implemented the solutions for these problems.
- Several companies and university labs make use of our data fusion tool DataIn to manage and analyze their production and experimental data.
- Our Control Loop Performance Monitoring tool APROMON has been used at numerous plants in various industries to monitor and analyse the data of hundreds of control loops and to report on the control quality.
- We have developed a soft sensor that predicts the moisture of paper at the end of the paper machine based on upstream process variables.
